DOCSLIB.ORG

Arxiv:1401.4655V1 [Cs.OH] 19 Jan 2014 Eoyssesi Plcto N Srbhvo Eedn 2.More [2]

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Arxiv:1401.4655V1 [Cs.OH] 19 Jan 2014 Eoyssesi Plcto N Srbhvo Eedn 2.More [2] The Energy/Frequency Convexity Rule: Modeling and Experimental Validation on Mobile Devices Karel De Vogeleer1, Gerard Memmi1, Pierre Jouvelot2, and Fabien Coelho2 1 TELECOM ParisTech – INFRES – CNRS LTCI - UMR 5141 – Paris, France 2 MINES ParisTech – CRI – Fontainebleau, France {karel.devogeleer,gerard.memmi}@telecom-paristech.fr, {pierre.jouvelot,fabien.coelho}@mines-paristech.fr Abstract. This paper provides both theoretical and experimental evi- dence for the existence of an Energy/Frequency Convexity Rule, which relates energy consumption and CPU frequency on mobile devices. We monitored a typical smartphone running a specific computing-intensive kernel of multiple nested loops written in C using a high-resolution power gauge. Data gathered during a week-long acquisition campaign suggest that energy consumed per input element is strongly correlated with CPU frequency, and, more interestingly, the curve exhibits a clear minimum over a 0.2 GHz to 1.6 GHz window. We provide and motivate an ana- lytical model for this behavior, which fits well with the data. Our work should be of clear interest to researchers focusing on energy usage and minimization for mobile devices, and provide new insights for optimiza- tion opportunities. Keywords: energy consumption and modeling, DVFS, power consump- tion, execution time modeling, smartphone, bit-reverse algorithm. 1 Introduction The service uptime of battery-powered devices, e.g., smartphones, is a sensitive issue for nearly any user [1]. Even though battery capacity and performance are hoped to increase steadily over time, improving the energy efficiency of current battery-powered systems is essential because users expect right now communica- tion devices to provide data access every time, everywhere to everyone. Under- standing the energy consumption of the different features of (battery-powered) arXiv:1401.4655v1 [cs.OH] 19 Jan 2014 computer systems is thus a key issue. Providing models for energy consumption can pave the way to energy optimization, by design and at run time. The power consumption of Central Processing Units (CPUs) and external memory systems is application and user behavior dependent [2]. Moreover, for cache-intensive and CPU-bound applications, or for specific Dynamic Voltage and Frequency Scaling (DVFS) settings, the CPU energy consumption may dom- inate the external memory consumption [3]. For example, Aaron and Carroll [2] 2 K. De Vogeleer, G. Memmi, P. Jouvelot, and F. Coelho showed that, for an embedded system running equake, vpr, and gzip from the SPEC CPU2000 benchmark suite, the CPU energy consumption exceeds the RAM memory consumption, whereas crafty and mcf from the same suite showed to be straining more energy from the device RAM memory. Providing an accurate model of energy consumption for embedded and, more generally, energy-limited devices such as mobile phones is of key import to both users and system designers. To reach that goal, our paper provides both theo- retical and first experimental evidence for the existence of an Energy/Frequency Convexity Rule, that relates energy consumption and CPU frequency on mobile devices. This convexity property seems to ensure the existence of an optimal frequency where energy usage is minimal. This existence claim is based on both theoretical and practical evidence. More specifically, we monitored a Samsung Galaxy SII smartphone running Gold- Rader’s Bit Reverse algorithm [4], a small kernel based on multiple nested loops written in C, with a high-resolution power gauge from Monsoon Solutions Inc. Data gathered during a week-long acquisition campaign suggest that energy consumed per input element is strongly correlated with CPU frequency and, more interestingly, that the corresponding curve exhibits a clear minimum over a 0.2 GHz to 1.6 GHz window. We also provide and motivate an analytical model of this behavior, which fits well with the data. Our work should be of clear interest to researchers focusing on energy usage and minimization on mobile devices, and provide new insights for optimization opportunities. The paper is organized as follows. Section 2 introduces the notions of en- ergy and power, and how these can be decomposed in different components on electronic devices. Section 3 describes the power measurement protocol and methodology driving our experiments, and the C benchmark we used. Section 4 introduces our CPU energy consumption model, and shows that it fits well with the data. Section 5 outlines the Energy/Frequency Rule derived from our experiment and modeling. Related work is surveyed in Section 6. We conclude and discuss future work in Section 7. 2 Power Usage in Computer Systems The total power Ptotal consumed by a computer system, including a CPU, may be separated into two components: Ptotal = Psystem + PCPU, where PCPU is consumed by the CPU itself and Psystem by the rest of system. In a battery- powered hand-held computer device Psystem may include the power needed to light the LCD display, to enable and maintain I/O devices (including memory), to keep sensors online (GPS, gyro-sensors etc.), and others. The power consumption PCPU of the CPU we focus on here can be divided into two parts: PCPU = Pdynamic+Pleak, where Pdynamic is the power consumed by the CPU during the switching activities of transistors during computation. Pleak is power originating from leakage effects inherent to silicon-based transistors, and is in essence not useful for the CPU’s purposes. Pdynamic may be split into the power Pshort lost when transistors briefly short-circuit during gate state changes The Energy/Frequency Convexity Rule 3 and Pcharge, needed to charge the gates’ capacitors: Pdynamic = Pshort +Pcharge. In 2 the literature Pcharge is usually [5] defined as αCfV , where α is a proportional constant indicating the percentage of the system that is active or switching, C the capacitance of the system, f the frequency at which the system is switching and V the voltage swing across C. Pshort originates during the toggling of a logic gate. During this switching, the transistors inside the gate may conduct simultaneously for a very short time, creating a direct path between VCC and the ground. Even though this peak current happens over a very small time interval, given current high clock frequencies and large amount of logic gates, the short-circuit current may be non-negligible. Quantifying Pshort is gate specific but it may be approximated by deeming it proportional to Pcharge. Thus the power Pdynamic stemming from the switching activities and the short-circuit currents in a CPU is thus Pcharge + 2 (η 1)Pcharge, i.e., η αCLfV , where η is a scaling factor representing the effects of− short-circuit power.· Pleak originates from leakage currents that flow between differently doped parts of a metal-oxide semiconductor field-effect transistor (MOSFET), the basic building block of CPUs. The energy in these currents are lost and do not con- tribute to the information that is held by the transistor. Some leakage currents are induced during the on or off -state of the transistor, or both. Six distinct sources of leakage are identified [6]. Despite the presence of multiple sources of leakage in MOSFET transistors, the sub-threshold leakage current, gate leakage, and band-to-band tunneling (BTBT) dominate the others for sub-100 nm tech- nologies [7]. Leakage current models, e.g., as incorporated in the BSIM [6] micro models, are accurate yet complex since they depend on multiple variables. More- over, Pleak fluctuates constantly as it also depends on the temperature of the sys- tem. Consequently Pleak cannot be considered a static part of the system’s power consumption. Given the different sources of power consumption in a MOSFET based CPU, the potal power can be rewritten as Ptotal = Psystem+Pleak+Pdynamic. The relationship between the power P (t) (Watts or Joules/s) and the energy E(∆t) (Joules) consumed by an electrical system over a time period ∆t is given by ∆t ∆t E(∆t)= P (t) dt = I(t) V (t) dt, (1) Z0 Z0 · where I(t) is the current supplied to the system, and V (t) the voltage drop over the system. Often V (t) is constant over time, hence dP (t)/dt only depends on I(t). If both current and voltage are constant over time, the energy integral becomes the product of voltage, current and time, or alternatively power and time. 3 Power Measurement Protocol on Mobile Devices A Samsung Galaxy S2 is used in our testbed sporting the Samsung Exynos 4 Systems-on-Chip (SoC) 45nm dual-core. The Galaxy S2 has a 32KB L1 data and instruction cache, and a 1 MB L2 cache. The mobile device runs Android 4 K. De Vogeleer, G. Memmi, P. Jouvelot, and F. Coelho 4.0.3 on the Siyah kernel adopting Linux 3.0.31. The frequency scaling governor in Linux was set to operate in userspace mode to prevent frequency and voltage scaling on-the-fly. The second CPU core was disabled during measurements. The smartphone is booted in clockwork recovery mode to minimize noisy side-effects of the Operating System (OS) and other frameworks. During the experiments, the phone’s battery was replaced by a power supply (Monsoon Power Monitor) that measures the power consumption at 5 kHz with an accuracy of 1mW. The power of the system and the temperature of the CPU were simultaneously logged. The kernel was patched to print a temperature sample to the kernel debug output at a rate of 2 Hz. The bit-reverse algorithm is used as benchmark kernel. This is an important operation since it is part of the ubiquitous Fast Fourier Transformation (FFT) algorithm, and rearranges deterministically elements in an array. The bit-reversal kernel is CPU intensive, induces cache effects, and is economically pertinent. The Gold-Rader implementation of the bit-reverse algorithm, often considered the reference implementation [4], is given below: void bitreverse_gold_rader (int N, complex *data) { int n = N, nm1 = n-1; int i = 0, j = 0; for (; i < nm1; i++) { int k = n >> 1; if (i < j) { complex temp = data[i]; data[i] = data[j]; data[j] = temp; } while (k <= j) {j -= k; k >>= 1;} j += k; } } The input of the bit-reversal algorithm is an array with a size of 2N ; the elements are pairs of 32 bit integers, representing complex numbers.
Load more

Recommended publications
  • Analysis of Body Bias Control Using Overhead Conditions for Real Time Systems: a Practical Approach∗
    IEICE TRANS. INF. & SYST., VOL.E101–D, NO.4 APRIL 2018 1116 PAPER Analysis of Body Bias Control Using Overhead Conditions for Real Time Systems: A Practical Approach∗ Carlos Cesar CORTES TORRES†a), Nonmember, Hayate OKUHARA†, Student Member, Nobuyuki YAMASAKI†, Member, and Hideharu AMANO†, Fellow SUMMARY In the past decade, real-time systems (RTSs), which must in RTSs. These techniques can improve energy efficiency; maintain time constraints to avoid catastrophic consequences, have been however, they often require a large amount of power since widely introduced into various embedded systems and Internet of Things they must control the supply voltages of the systems. (IoTs). The RTSs are required to be energy efficient as they are used in embedded devices in which battery life is important. In this study, we in- Body bias (BB) control is another solution that can im- vestigated the RTS energy efficiency by analyzing the ability of body bias prove RTS energy efficiency as it can manage the tradeoff (BB) in providing a satisfying tradeoff between performance and energy. between power leakage and performance without affecting We propose a practical and realistic model that includes the BB energy and the power supply [4], [5].Itseffect is further endorsed when timing overhead in addition to idle region analysis. This study was con- ducted using accurate parameters extracted from a real chip using silicon systems are enabled with silicon on thin box (SOTB) tech- on thin box (SOTB) technology. By using the BB control based on the nology [6], which is a novel and advanced fully depleted sili- proposed model, about 34% energy reduction was achieved.
    [Show full text]
  • Dynamic Voltage/Frequency Scaling and Power-Gating of Network-On-Chip with Machine Learning
    Dynamic Voltage/Frequency Scaling and Power-Gating of Network-on-Chip with Machine Learning A thesis presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Master of Science Mark A. Clark May 2019 © 2019 Mark A. Clark. All Rights Reserved. 2 This thesis titled Dynamic Voltage/Frequency Scaling and Power-Gating of Network-on-Chip with Machine Learning by MARK A. CLARK has been approved for the School of Electrical Engineering and Computer Science and the Russ College of Engineering and Technology by Avinash Karanth Professor of Electrical Engineering and Computer Science Dennis Irwin Dean, Russ College of Engineering and Technology 3 Abstract CLARK, MARK A., M.S., May 2019, Electrical Engineering Dynamic Voltage/Frequency Scaling and Power-Gating of Network-on-Chip with Machine Learning (89 pp.) Director of Thesis: Avinash Karanth Network-on-chip (NoC) continues to be the preferred communication fabric in multicore and manycore architectures as the NoC seamlessly blends the resource efficiency of the bus with the parallelization of the crossbar. However, without adaptable power management the NoC suffers from excessive static power consumption at higher core counts. Static power consumption will increase proportionally as the size of the NoC increases to accommodate higher core counts in the future. NoC also suffers from excessive dynamic energy as traffic loads fluctuate throughout the execution of an application. Power- gating (PG) and Dynamic Voltage and Frequency Scaling (DVFS) are two highly effective techniques proposed in literature to reduce static power and dynamic energy in the NoC respectively.
    [Show full text]
  • Real-Time Dynamic Voltage Scaling for Low-Power Embedded Operating Systems£
    Real-Time Dynamic Voltage Scaling for Low-Power Embedded Operating Systems£ Padmanabhan Pillai and Kang G. Shin Real-Time Computing Laboratory Department of Electrical Engineering and Computer Science The University of Michigan Ann Arbor, MI 48109-2122, U.S.A. pillai,kgshin @eecs.umich.edu ABSTRACT ful microprocessors running sophisticated, intelligent control soft- In recent years, there has been a rapid and wide spread of non- ware in a vast array of devices including digital camcorders, cellu- traditional computing platforms, especially mobile and portable com- lar phones, and portable medical devices. puting devices. As applications become increasingly sophisticated and processing power increases, the most serious limitation on these Unfortunately, there is an inherent conflict in the design goals be- devices is the available battery life. Dynamic Voltage Scaling (DVS) hind these devices: as mobile systems, they should be designed to has been a key technique in exploiting the hardware characteristics maximize battery life, but as intelligent devices, they need powerful of processors to reduce energy dissipation by lowering the supply processors, which consume more energy than those in simpler de- voltage and operating frequency. The DVS algorithms are shown to vices, thus reducing battery life. In spite of continuous advances in be able to make dramatic energy savings while providing the nec- semiconductor and battery technologies that allow microprocessors essary peak computation power in general-purpose systems. How- to provide much greater computation per unit of energy and longer ever, for a large class of applications in embedded real-time sys- total battery life, the fundamental tradeoff between performance tems like cellular phones and camcorders, the variable operating and battery life remains critically important.
    [Show full text]
  • Learning-Directed Dynamic Voltage and Frequency Scaling Scheme with Adjustable Performance for Single-Core and Multi-Core Embedded and Mobile Systems †
    sensors Article Learning-Directed Dynamic Voltage and Frequency Scaling Scheme with Adjustable Performance for Single-Core and Multi-Core Embedded and Mobile Systems † Yen-Lin Chen 1,* , Ming-Feng Chang 2, Chao-Wei Yu 1 , Xiu-Zhi Chen 1 and Wen-Yew Liang 1 1 Department of Computer Science and Information Engineering, National Taipei University of Technology, Taipei 10608, Taiwan; [email protected] (C.-W.Y.); [email protected] (X.-Z.C.); [email protected] (W.-Y.L.) 2 MediaTek Inc., Hsinchu 30078, Taiwan; [email protected] * Correspondence: [email protected]; Tel.: +886-2-27712171 (ext. 4239) † This paper is an expanded version of “Learning-Directed Dynamic Volt-age and Frequency Scaling for Computation Time Prediction” published in Proceedings of 2011 IEEE 10th International Conference on Trust, Security and Privacy in Computing and Communications, Changsha, China, 16–18 November 2011. Received: 6 August 2018; Accepted: 8 September 2018; Published: 12 September 2018 Abstract: Dynamic voltage and frequency scaling (DVFS) is a well-known method for saving energy consumption. Several DVFS studies have applied learning-based methods to implement the DVFS prediction model instead of complicated mathematical models. This paper proposes a lightweight learning-directed DVFS method that involves using counter propagation networks to sense and classify the task behavior and predict the best voltage/frequency setting for the system. An intelligent adjustment mechanism for performance is also provided to users under various performance requirements. The comparative experimental results of the proposed algorithms and other competitive techniques are evaluated on the NVIDIA JETSON Tegra K1 multicore platform and Intel PXA270 embedded platforms.
    [Show full text]
  • Energy Proportional Computing in Commercial Fpgas with Adaptive
    Energy proportional computing in Commercial FPGAs with Adaptive Voltage Scaling Jose Nunez-Yanez Department of Electrical and Electronic Engineering University of Bristol, UK +44 117 3315128 [email protected] ABSTRACT an open-loop configuration. However, worst case variability is Voltage and frequency adaptation can be used to create energy rarely the case. For this reason, this paper investigates Adaptive proportional systems in which energy usage adapts to the amount Voltage Scaling (AVS) in which run-time monitoring of of work to be done in the available time. Closed-loop voltage and performance variability in the silicon is used together with system frequency scaling can also take into account process and characterization to influence the voltage and the frequency on the temperature variations in addition to system load and this removes fly in a closed-loop configuration. a significant proportion of the margins used by device The rest of the paper is structured as follows. Section 2 describes manufacturers. This paper explores the capabilities of commercial related work. Section 3 presents the hardware platform used in FPGAs to use closed-loop adaptive voltage scaling to improve this research while section 4 introduces the design flow that their energy and performance profiles beyond nominal. An embeds the AVS capabilities in the user design. Section 5 adaptive power architecture based on a modified design flow is presents the power adaptive architecture based on the novel in- created with in-situ detectors and dynamic reconfiguration of situ detectors. Section 6 presents and discusses the results clock management resources. The results of deploying AVS in focusing on power and energy measurements.
    [Show full text]
  • The Dynamic Voltage and Frequency Scaling Based on the On-Chip Microcontroller System
    Journal of Theoretical and Applied Information Technology 10th May 2013. Vol. 51 No.1 © 2005 - 2013 JATIT & LLS. All rights reserved. ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 THE DYNAMIC VOLTAGE AND FREQUENCY SCALING BASED ON THE ON-CHIP MICROCONTROLLER SYSTEM 1,2TIEFENG LI, 1CAIWEN MA, 1WENHUA LI 1 Xi'an Institute of Optics and Precision Mechanics of Chinese Academy of Sciences, Xi’an, 710119,China 2The Graduate University of Chinese Academy of Sciences ,Beijing, 100049, China E-mail: [email protected] , [email protected] , [email protected] ABSTRACT With the rapid increase of complexity and size of the on-chip microcontroller system (OCMS), the power consumption issue for the OMCS is increasingly becoming critical and needs to be solved quickly. Because the CPU is often the major power consumer in the OCMS, so an important strategy to achieve energy saving is via the dynamic voltage and frequency scaling (DVFS), which can enable a processor to operate at a range of voltages and frequencies. However, it needs to be emphasized that the conventional DVFS is fully executed by the software scheduler in the operating system, and its main drawback is that the scheduler can’t accurately track the performance requirements of CPU when the dormant frequency of CPU is increasing continuously. In this paper, we firstly present a typical hardware DVFS architecture, which can automatically carry out DVFS without the software scheduler involvement. Therefore, it avoids increasing the software's workload and reduces the power consumption. Keywords: Power Consumption, DVFS, Software Scheduler, Performance Requirement 1.
    [Show full text]
  • LG8: (E) Engineering and Physical Considerations
    LG8: (E) Engineering and Physical Considerations Topics: Power consumption, scaling, size, logical effort and performance limits. LG8.1 - E - 90 Nanometer Gate Length. LG8.2 - E - Power Consumption LG8.3 - E - Dynamic Power Gating LG8.4 - E - Dynamic Frequency Scaling LG8.5 - E - Dynamic Voltage Scaling LG8.6 - E - Logical Effort LG8.7 - E - Information Flux What is the Silicon End Point ? 1 LG8.1 - E - 90 Nanometer Gate Length. The mainstream VLSI technology in the period 2004-2008. Parameters from a 90 nanometer standard cell library: Parameter Value Unit Drawn Gate Length 0.08 µm Metal Layers 6 to 9 layers Max Gate Density 400K gates/mm2 Track Width 0.25 µm Track Spacing 0.25 µm Tracking Capacitance 1 fF/mm Core Supply Voltage 0.9 to 1.4 V FO4 Delay 51 ps Typical processor core: 200k gates + 4 RAMs: one square millimeter. Typical SoC chip area is 50-100 mm2 20-40 million gates. Actual gate and transistor count higher owing to custom blocks (RAMs mainly). Now the industry is moving to 45 nanometer. 2007: Dual-core Intel Itanium2: 2 billion transistors. http://en.wikipedia.org/wiki/Moore’s_law 2 LG8.2- E - Power Consumption P = V × I = E × f I (current) = Static Current + Dynamic Current. Early CMOS (VCC= 5 volts): negligible static current. Today it’s 30 % of dynamic. Dynamic current = Short circuit current + Charge current. Charge current: • All energy in a net/gates is wasted each time it toggles. • The energy in a capacitor is E = CV 2/2. • Dominant capacitance is proportional to net length.
    [Show full text]
  • Analysis of Clocked Timing Elements for Dynamic Voltage Scaling Effects Over Process Parameter Variation Hoang Q
    Analysis of Clocked Timing Elements for Dynamic Voltage Scaling Effects over Process Parameter Variation Hoang Q. Dao Kevin Nowka Vojin G. Oklobdzija ACSEL Lab IBM Austin Research Lab ACSEL Lab University of California, Davis 11400 Burnet Road MS9460, University of California, Davis 1 Shields Avenue, Davis, CA 95616 Austin, TX 78758 1 Shields Avenue, Davis, CA 95616 +1-530-754-6827 +1-512-838-3350 +1-530-754-6827 [email protected] [email protected] [email protected] ABSTRACT It was our interest to analyze the effect of large supply voltage In power-constrained systems, the power efficiency of latches and variation as well as the typical process and environmental flip-flops is pivotal. Characteristics of three selected latches and variation in bulk CMOS circuits. In assessing the quality of FFs were analyzed for their behavior under voltage scaling and designs for the low power, both the designs and the evaluation different process corners in a 0.18um CMOS technology. The methodology and criteria are equally important. We specifically relative performance amongst the latches/FFs was consistent focus on latches and flip-flops: an improved semi-dynamic flip- across the different supply voltages. At low-voltage power-delay- flop (SDFF) [1], the sense-amplifier flip-flop (SAFF) [2] and the product was degraded by about 25%. Energy-delay-product was master-slave PowerPC latch [5]. Section 2 presents the effects of approximately doubled at low-voltage – for all latches/FFs over voltage scaling. Section 3 discusses modification to the test all process corners. This result was smaller in comparison to the bench by Stojanovic and Oklobdzija [3].
    [Show full text]
  • Dynamic Voltage and Frequency Scaling : the Laws of Diminishing Returns Authors Etienne Le Sueur and Gernot Heiser
    Dynamic Voltage and Frequency Scaling : The laws of diminishing returns Authors Etienne Le Sueur and Gernot Heiser Workshop on Power Aware Computing and Systems, pp. 1–5, Vancouver, Canada, October, 2010 1. Senior Member of Technical Staff at Vmware. During the research he was at NICTA is Australia's Information and Communications Technology Presented By Research Centre of Excellence. 2. Scientia Professor and Prasanth B L John Lions Chair of Operating Systems Aakash Arora School of Computer science and Engineering UNSW , Sydney, Australia Introduction • What contributions are in this paper (according to the authors)? The Authors analyzed and examined the potential of DVFS across three platforms with recent generation of AMD processors in various aspects viz.,. Scaling of Silicon Transistor technology, Improved memory performance, Improved sleep/ idle mode, Multicore Processors. • What possible consequences can the contributions have? The results shows that on the most recent platform, the effectiveness of DVFS is markedly reduced, and actual savings are only observed when shorter executions (at higher frequency) are padded with the energy consumed when idle. Previous Works Before Author related to DVFS Previous research has attempted to leverage DVFS as a means to improve energy efficiency by lowering the CPU frequency when cycles are being wasted, stalled on memory resources. Energy can only be saved if the power consumption is reduced enough to cover the extra time it takes to run the work load at the lower frequency. "Hot" gigahertz Higher the frequency more the power consumption • An example to understand the scaling of Voltage and frequency by simple instruction execution Case 1: Equal execution time for each stage Case 2: Unequal execution time for each stage Previous Research works Referenced They used simulated execution traces and the level of slack time to choose a new CPU frequency at each OS scheduler invocation [7].
    [Show full text]
  • Adaptive, Transparent Frequency and Voltage Scaling of Communication
    Adaptive, Transparent Frequency and Voltage Scaling of Communication Phases in MPI Programs ¡ Min Yeol Lim Vincent W. Freeh David K. Lowenthal Abstract because power is proportional to the product of the frequency and the square of the voltage. As an ex- Although users of high-performance computing are ample of the problem that is faced, several years ago most interested in raw performance, both energy and it was observed that on their current trend, the power power consumption have become critical concerns. density of a microprocessor will reach that of a nu- Some microprocessors allow frequency and voltage clear reactor by the year 2010 [17]. scaling, which enables a system to reduce CPU per- To balance the concerns of power and perfor- formance and power when the CPU is not on the crit- mance, new architectures have aggressive power ical path. When properly directed, such dynamic fre- controls. One common mechanism on newer micro- quency and voltage scaling can produce significant processors is the ability of the application or operat- energy savings with little performance penalty. ing system to select the frequency and voltage on the This paper presents an MPI runtime system that fly. We call this dynamic voltage and frequency scal- dynamically reduces CPU performance during com- ing (DVFS) and denote each possible combination of munication phases in MPI programs. It dynamically voltage and frequency a processor state, or p-state. identifies such phases and, without profiling or train- While changing p-states has broad utility, includ- ing, selects the CPU frequency in order to minimize ing extending battery life in small devices, the pri- energy-delay product.
    [Show full text]
  • Energy Consumption Evaluation of Flip-Flops for Dynamic Voltage Scaling Systems and Circulating-Temperature Applications
    Title Page Energy Consumption Evaluation of Flip-Flops for Dynamic Voltage Scaling Systems and Circulating-Temperature Applications by Minghe Shao Submitted to the Graduate Faculty of the Swanson School of Engineering in partial fulfillment of the requirements for the degree of Master of Science University of Pittsburgh 2021 Committee Membership Page UNIVERSITY OF PITTSBURGH SWANSON SCHOOL OF ENGINEERING This thesis was presented by Minghe Shao It was defended on March 5, 2021 and approved by In Hee Lee, PhD, Assistant Professor, Department of Electrical and Computer Engineering Samuel Dickerson, PhD, Assistant Professor, Department of Electrical and Computer Engineering Rajkumar Kubendran, PhD, Assistant Professor, Department of Electrical and Computer Engineering Thesis Advisor: In Hee Lee, PhD, Assistant Professor, Department of Electrical and Computer Engineering ii Copyright © by Minghe Shao 2021 iii Abstract Energy Consumption Evaluation of Flip-Flops for Dynamic Voltage Scaling Systems and Circulating-Temperature Applications Minghe Shao, M.S. University of Pittsburgh, 2021 CMOS circuit technology has developed with a help of transistor scaling. In past decades, previous studies found that operating environment of digital system affects circuit performance significantly. For example, extensive research has been performed to learn temperature dependency of CMOS circuits and optimize their performance at given temperature. Adaptive voltage scaling (AVS) is one of the important techniques, responses to operating temperature and dynamically change supply voltage of digital circuits to optimize circuit performance. AVS can enable energy optimization for a system experiencing significant temperature variation, including a space satellite. However, inappropriate AVS results in functional failure in an extreme condition, which can cause substantial problem for a critical mission.
    [Show full text]
  • Power Management and Dynamic Voltage Scaling: Myths and Facts
    Power Management and Dynamic Voltage Scaling: Myths and Facts David Snowdon, Sergio Ruocco and Gernot Heiser National ICT Australia and School of Computer Science and Engineering University of NSW, Sydney 2052, Australia [email protected] Abstract frequency. The total energy ¨ for the computation is then proportional to the square of the voltage: This paper investigates the validity of common ap- ¤ ¨ ¢ ¦ proaches to power management based on dynamic volt- ¥ (2) age scaling (DVS). Using instrumented hardware and ap- propriate operating-system support, we account separately Note that the total energy for a computation does in this for energy consumed by the processor and the memory sys- simple model not depend on the frequency, but a reduced tem. We find that memory often contributes significantly to core voltage requires a reduction of the clock frequency and overall power consumption, which leads to a much more therefore implies a longer overall execution time. complex relationship between energy consumption and core The assumptions behind Eqn. 2 are highly dubious, as voltage and frequency than is frequently assumed. As a they ignore other system components, in particular the consequence, we find that the voltage and frequency setting front-side bus and memory [2]. Those other components that minimises energy consumption is dependent on system impact the execution time of a program, leading to a much characteristics, and, more importantly, on the application- more complex dependence on the processor frequency. Fur- specific balance of memory and CPU activity. The optimal thermore, those components themselves consume energy, setting of core voltage and frequency therefore requires ei- and that energy consumption scales differently than the pro- ther a-priori analysis of the application or, where this is not cessor’s.
    [Show full text]